Heapsort

Note: Heapsort is not a divide-and-conquer algorithms.
      Instead, it is an algorithm based on a
      data structure, a max-heap.


1) Define the "binary tree" data structure.

2) A "nearly complete binary tree" is a binary tree in
   which every level of the tree, except possibly the
   bottom level, is completely filled in and all the nodes
   in the bottom level are as far to the left as possible.

3) A nearly complete binary tree can be stored efficiently
   in an array (in breadth-first order).

   Let i be the index in the array of any node of the binary
   tree. Then its parent, left child, and right child nodes
   have the following indices in the array.
      parent(i) = (i - 1) / 2   // integer division
      left(i)  = 2 * i + 1
      right(i) = 2 * i + 2

4) Every array can be thought of as representing a nearly
   complete binary tree.

5) A "max-heap" is a nearly complete binary tree in which
   the value stored in every node is at least as large
   as the values in its two child nodes.

6) Since a max-heap is a nearly complete binary tree,
   we can, if we want, store a max-heap in an array.

7) An array that holds a max-heap is not sorted, but it
   is "kind of" sorted because of the way a max-heap
   is organized.

8) The heapify() method takes a nearly complete binary
   tree in which each child of the root is a max-heap,
   and converts the binary tree into a max-heap.

   You should visualize the heapify() method as acting
   on a real binary tree, not the array storage of the
   binary tree.

9) You convert an arbitrary array into a max-heap by starting
   at the last parent node and then heapifying every sub tree
   all the way up to the root.

   Again, you are better off visualizing this algorithm as acting
   on a real btree, rather than on the array storage of the btree.

10) You sort an array of length n that holds a max-heap by
   a) swapping the last element of the max-heap
      with the root of the max-heap,
   b) decrease the size of the max-heap by 1,
   c) re-heapifying the (now smaller) nearly
      complete binary tree,
   d) repeat n-1 times

11) The heapsort algorithm starts with an arbitrary array, then
     uses (9) to convert the array into a max-heap, then uses (10)
     to sort the max-heap.